ggml-org/llama.cpp

Pull Requests Commits

clean up logging and timing

0cc4m committed 20 days ago

42874dfd

simplify replay submission

0cc4m committed 20 days ago

71d9373b

fix queue debug utils label

0cc4m committed 20 days ago

f10a92dd

Revert "submit only twice for graph reuse"

0cc4m committed 25 days ago

9c1d7406

submit only twice for graph reuse

0cc4m committed 25 days ago

e218a390

vulkan: capture and replay command buffers where possible

0cc4m committed 25 days ago

ccceabc0

ui: PWA support (#23871)

allozaur committed 25 days ago

Verified f7ca93d1

fit : avoid including llama-ext.h in fit.h (#24506)

ggerganov committed 25 days ago

Verified 02182fc5

ggerganov committed 25 days ago

f532be8f

ggml : bump version to 0.15.1 (ggml/1541)

ggerganov committed 25 days ago

e08c226a

vendor : update cpp-httplib to 0.47.0 (#24395)

angt committed 25 days ago

Verified 70b54e14

UI/jpeg exif orientation (#24196)

ServeurpersoCom committed 25 days ago

Verified 6471e3c0

spec: add EAGLE3 speculative decoding support (#18039)

ruixiang63 committed 25 days ago

Verified 88a39274

ggml: support concat for scalar types at cuda backend (#24011)

zihaomu committed 25 days ago

Verified 85f99dca

[SYCL] Fix CI build & release for SYCL backend (#24387)

arthw committed 25 days ago

Verified 099ea76f

opencl: add q5_0/q5_1 gemm and gemv kernels for Adreno (#24319)

shaofeiqi committed 25 days ago

Verified ba1df050

docker : support specifying the GCC version for CUDA (#24447)

wencan committed 26 days ago

Verified 1593d568

vulkan: ifdef eMesaHoneykrisp (build fix) (#24479)

jeffbolznv committed 26 days ago

Verified 4c659550

ggerganov committed 26 days ago

263cc04a

ggml : bump version to 0.15.0 (ggml/1539)

ggerganov committed 26 days ago

17e59d62

vulkan: add fast path for contiguous buffer transfers (#23973)

winstonma committed 26 days ago

Verified fdc3db9b

vulkan: use medium matmul tile on Asahi Linux (#24306)

xingjianll committed 26 days ago

Verified 1af154a7

server: skip unused log lines on router mode (#24463)

ngxson committed 26 days ago

Verified 18ef86ec

vocab : adopt leading TemplateProcessing special token as BOS (#24428)

o7si committed 26 days ago

Verified 1bfbdb13

vocab : refactor normalizer flags into options struct, add strip_accents (#24371)

o7si committed 26 days ago

Verified 68f30663

server : skip checkpoints beyond pos_next (#24411)

aldehir committed 26 days ago

Verified db94854f

vendor : update LibreSSL to 4.3.2 (#24397)

angt committed 27 days ago

Verified ac4cddeb

Remove padding and multiple D2D copies for MTP (#24086)

gaugarg-nv committed 27 days ago

Verified e95dae18

chat: fix LFM2/LFM2.5 ignoring json_schema (#24377)

tdakhran committed 27 days ago

Verified d2462f8f

CUDA: Fix ssm_scan_f32 data-races (#24360)

ORippler committed 27 days ago

Verified fb83cc9a

Older